Deep.Net


GaussianProcess

Namespace: Models

Nested types and modules

TypeDescription
HyperPars

Gaussian Process hyperparameter type.

Kernel

Kernel Type.

Pars

Parameter Type of a Gaussian Process dependent on the used Kernel.

ParsLinear

Gaussian Process parameters with linear kernel.

ParsSE

Gaussian Process parameters with squared exponential kernel.

Functions and values

Function or valueDescription
defaultHyperPars
Signature: HyperPars

The dafault hyperparameters.

initLengthscale l seed shp
Signature: l:single -> seed:'?175765 -> shp:int64 list -> Tensor<single>
Type parameters: '?175765

Iitializes the lengthscale.

initSignalVariance s seed shp
Signature: s:single -> seed:'?175767 -> shp:int64 list -> Tensor<single>
Type parameters: '?175767

Initializes the signal variance.

linearCovariance x y
Signature: x:ExprT -> y:ExprT -> ExprT

Calculates covariance matrix between two vectors using linear kernel.

logMarginalLiklihood (...)
Signature: pars:Pars -> x:'?175779 -> y:'?175780 -> sigmaNs:'?175781 -> xStar:'?175782 -> '?175783
Type parameters: '?175779, '?175780, '?175781, '?175782, '?175783

WARNING: NOT YET IMPLEMENTED, ONLY A REMINDER FOR LATER IMPLEMENTATION! !!! CALLING THIS FUNCTION WILL ONLY CAUSE AN ERROR !!!

pars mb hp
Signature: mb:ModelBuilder<single> -> hp:HyperPars -> Pars

Parameters of the Gaussian Process.

predict pars x y sigmaNs xStar
Signature: pars:Pars -> x:ExprT -> y:ExprT -> sigmaNs:ExprT -> xStar:ExprT -> ExprT * ExprT
squaredExpCovariance (l, sigf) x y
Signature: (l:ExprT * sigf:ExprT) -> x:ExprT -> y:ExprT -> ExprT

Calculates covariance matrix between two vectors using linear kernel.

Fork me on GitHub